A bloke I worked with took a promotion to Adelaide about six years ago. Nothing went wrong. The job was good, the house was better than the one he left, and his wife found work inside a month. I saw him at a conference two years later and asked who he had ended up spending time with down there, and he thought about it for slightly too long before saying "work people, mostly".

I have thought about that pause more than the answer. He had not lost anything dramatic. He had simply arrived somewhere with a full house and an empty Saturday, and nothing about that is a story anybody tells at the time.

What I went looking for, and did not find

I wanted the study that follows men who relocate for work and counts what happens to their friendships. Compares them against men who stayed. Tells you whether the circle shrinks, and how long it takes to come back if it does.

It does not exist, as far as I can find. There is a large migration literature about wages and employment, and a large loneliness literature about age and partnership, and the two have not been introduced. Nobody has run the study that would let me tell you what a work move costs socially.

That absence is worth stating out loud, because plenty of writing on this subject speaks with confidence it has not earned. If someone tells you a relocation shrinks your circle by a measurable amount, ask them where the number came from. I could not find one.

What follows is what the evidence does support, which is narrower and still useful.

The one venue a work move actually closes

This part is measurable, and it is the sharpest thing in the article.

Work is the single largest source of close friends anybody has counted. The Survey Center on American Life's State of American Friendship, an Ipsos KnowledgePanel probability survey of 2,019 US adults fielded 14 to 23 May 2021, asked people with close friends where they had met one. Fifty-four per cent named their own workplace or their spouse's. School came second at 47 per cent, existing friends at 40, the neighbourhood at 35.

Two things about that number before it gets stretched, both of which matter here.

It is all adults, not men. The report does not break workplace friendship formation down by sex at all, so anyone printing it as a fact about men has added the word.

And it is a share of people, not of friendships. Fifty-four per cent of people with close friends met at least one of them at a workplace. The options were plainly not exclusive, since they sum to well over a hundred. So the honest reading is that work is named by more people than any other venue, not that most friendships start there.

Even read that carefully, the implication for a man who moves holds. The venue named by more people than any other has just been replaced with a different building full of strangers, and the friendships that venue produced are now interstate.

What the Australian data does attach loneliness to

If the relocation question is unanswered, the neighbouring one is answered well.

Botha and Bower published an analysis in BMC Public Health on 10 May 2024 using HILDA, the Australian household panel. Waves 2 to 21, covering 2002 to 2021: 118,667 person-year observations from 12,117 Australian men aged 15 to 98, analysed with linear fixed effects, which follows the same men over time rather than comparing different men at different ages.

What predicts a lonelier year for an Australian man, in their data:

  • Having a partner is protective at every age, worth between 0.32 and 0.81 points on a seven-point scale.
  • Being employed reduces loneliness.
  • Couples with children report higher loneliness in middle age than couples without them.
  • Loneliness peaks twice across a life: once in young adulthood, once in the late forties, with the lowest levels around seventy.

The finding I keep coming back to is a null. The study found no significant association between loneliness and residential region, rural against urban.

That one is easy to stretch, so I want to quote it precisely. It says where a man lives does not predict how lonely he is. It says nothing at all about moving, because relocation is not something the study tested. A man in Adelaide is not lonelier than a man in Melbourne on this data. Whether a man who went from one to the other is lonelier than he was, nobody has measured.

Those are different claims and the gap between them is exactly where the confident writing lives.

The thing that actually moved

Here is what I think happened to my colleague, offered as a reading rather than a finding.

Nothing in his life got worse. His partner came with him, so the most protective variable in the Australian data travelled intact. He stayed employed, so the second one did too. On every measure Botha and Bower could see, he should have been fine, and by his own account he mostly was.

What he lost was not a variable anybody counts. It was the accumulated, unarranged contact of twenty years in one city: the bloke two streets over, the five-a-side he had been in since his thirties, the people whose birthdays he knew without checking. None of that is partnership, employment or region. It is the layer underneath, and it is the layer the panel data has no column for.

That is my best explanation for why the research reads as reassuring and the experience does not.

What has to be rebuilt, and in what order

The instinct is to treat this as a friendship problem and go looking for friends, which is the slowest possible entry point. The order that seems to work is duller.

First, a fixed thing on a fixed day. Not a person, a slot. The reason is mechanical: a recurring commitment produces repeated contact with the same faces without anybody having to arrange it, which is the property the old city supplied for free. I have written about why the activity matters more than the people at the start.

Second, accept the conversion rate is slow. Six weeks of turning up buys recognition and very little else, and men who do not know that in advance read the ordinary rate as personal failure.

Third, keep one line open to the old city. This is the part men skip, and it is the cheapest of the three. The friendships back home do not need replacing; they need a message every few months to stop them going quiet, and restarting one that already went quiet is harder than not letting it.

My colleague did the first of those about four years in, through his kids' cricket rather than by design. He describes it now as the point things turned, which is a longer run at it than anybody would choose.

Sources

  • Survey Center on American Life, The State of American Friendship: Change, Challenges, and Loss (2021). Ipsos KnowledgePanel, n=2,019 US adults, fielded 14-23 May 2021, margin of error ±2.4 points. Venue figures are all adults, not men, and are shares of people rather than of friendships. https://www.americansurveycenter.org/research/the-state-of-american-friendship-change-challenges-and-loss/
  • Botha F, Bower M. Predictors of male loneliness across life stages: an Australian study of longitudinal data. BMC Public Health, 10 May 2024. HILDA waves 2-21 (2002-2021), 118,667 person-year observations from 12,117 men aged 15-98, linear fixed effects. Observational panel data: these are associations within men over time, not effects. https://pubmed.ncbi.nlm.nih.gov/38730388/